I Sold ETH Covered Calls in a DeFi Options Vault for 60 Days: Premiums, Assignment, and What 8 ETH Actually Earned

In July 2026, ETH washed out to roughly $1,600 and I did nothing but watch. When it clawed back about 65% and camped between $2,600 and $2,800 through August, my 8 ETH — about $21,600 — felt like dead weight. So I did the thing every dashboard dangles: I deposited into a covered call vault on Derive and sold weekly out-of-the-money calls for eight straight epochs.

Sixty days later the ledger read: $823 in premiums, $411 forfeited to one assignment, about $26 in fees — roughly 1.9% net in two months, against a front page advertising up to 30% APY. This is that ledger in full, plus the hack that changed how I vet these contracts. Education, not financial advice — every number is my own fill or dated public data.

What a covered call vault actually does

The mechanics are older than DeFi. You own ETH; the vault sells call options against it weekly; you keep the premium. Below the strike at expiry, the option dies and the premium is pure income; above it, your ETH is sold at the strike and you keep only the premium. You are trading upside for rent.

On-chain versions are stricter. Derive and Aevo vaults sell European options that settle once, at expiry, against an oracle price — Derive uses Chainlink — with no early close and no mid-week rolls. Epochs add a second constraint: deposits lock for the week and withdrawals process at epoch end. The income is real but smaller than banners suggest: at ETH around $2,700 with implied vol of 45–60%, 5–10% out-of-the-money weeklies grossed roughly 0.3–1.0% of notional per week — 15–50% annualized gross. That is where the marketing stops; fees, assignment gaps and flat weeks live downstream. Compounding premiums follow the same math as any auto-compounding vault, and the headline rate obeys the same trap as APY versus APR.

The setup: 8 ETH into Derive in August 2026

Context for the entry. ETH peaked above $4,900 in 2025, fell to about $1,600 in July 2026, then recovered roughly 65% into the $2,600–2,800 band, with trailing one-year realized volatility swinging between 35% and 79%. After a washout like that I wanted to get paid while undecided on direction — the textbook covered-call regime. I moved the coins per our exchange-to-DeFi walkthrough, with bridge hygiene from the bridge safety guide.

I picked Derive for boring reasons: the ex-Lyra stack is the clear leader with roughly 95% of on-chain options premium volume, TVL of $198M on September 24, 2026 (up 26% in 30 days), a DRV listing on Coinbase in May 2026, and a V3 zero-knowledge upgrade announced in September. Leader status is not safety — Ribbon proves that — but depth matters when you are short options. I read the vault's admin functions and oracle wiring against our audit checklist before moving a satoshi.

Eight epochs, one table

The complete run. Strikes sat 6–11% out of the money at sale; premiums are the vault's weekly credit as a percentage of notional.

*Week 8 ran on 8.06 ETH re-entered at $3,060 after the week-7 assignment — the tax paid in units, not dollars.

Six quiet weeks: sideways chop, every strike expiring worthless, premiums compounding. Then came week 7.

The assignment week, dollar by dollar

On Monday of week 7, ETH sat near $2,760 and implied vol had cooled toward 40%, so the $3,000 strike — about 9% out of the money — paid just 0.35%, the cheapest premium of the run. By Friday, ETH closed at $3,050, a 10% weekly run straight through my strike.

The math, precisely: the position had compounded to about 8.22 ETH, worth $25,071 at the close. Assignment sold it at $3,000, forfeiting $50 per ETH — $411, or 1.6% of the position — while keeping the $81 premium; net versus holding, minus $330. Re-entering at $3,060 turned 8.22 ETH into 8.06. That shrinkage is the assignment tax in units, and it is permanent. This is not a malfunction — it is the trade: selling calls means selling the top on purpose, and my mistake was violating the rule the best research keeps repeating.

The 30% APY label versus the 60-day ledger

The front page advertised up to 30% APY; the trailing-30-day dashboard showed 17.4%. My ledger says:

• Gross premiums: $823 — 3.76% of notional over 60 days, ~23% annualized.
• Assignment gap: −$411 (week 7).
• Fees: −$26.
• Net: about $386 — 1.8% over 60 days, ~11% annualized after everything; 1.9% before fees.

So even the gross run-rate (23%) never touched the label (30%), and the net number was about a third of the dashboard's implication. Anchorage Digital's May 28, 2026 whitepaper found unfiltered 20-delta, 30-day BTC calls netted only 5.5% yield in the twelve months to April 2026, with regime filtering adding 5.2% annualized from October 2021 to April 2026. Their productive corridor — 10–25 delta, expiries of at least 21 days, in a market where crypto's volatility risk premium runs 2–3 times richer than SPY or QQQ — is exactly where my 7-day weeklies at 6–11% OTM did not sit. Same lesson as our DeFi yield reality check; the BTC version's numbers are in our Bitcoin yield piece.

Fees and friction: the boring line items

Derive charges 0.1–0.3% on trades plus 0.1% on exercise and settlement; eight epochs cost me about $26. Gas was a non-event — Derive and Aevo run on app-chains where transactions cost fractions of a cent, and ethereum.org's September 17, 2026 data puts a token swap at $1.20–4.50 on Ethereum mainnet, $0.001–0.002 on Base and $0.005–0.008 on Arbitrum. Deposit on the vault's own chain and gas never decides this trade, though our gas fees versus yield breakdown covers when it does.

The real friction is temporal: withdrawals queue for epoch end, my exit added an idle weekend, and the position sat out a re-entry cycle after assignment — roughly five of the 60 days earned nothing. That drag belongs in your model, as does the tax reality that premiums are income the moment they land. Queue mechanics mirror our withdrawal queues guide; reporting is in the 2026 DeFi tax guide.

The October 2026 on-chain options scorecard

Where I parked money is the surviving half of a graveyard:

Derive's dominance — more than $30B lifetime notional, with TVL across Hyperliquid L1 ($59.6M), Base ($50.9M), Ethereum ($34.4M), OP Mainnet ($32.8M) and Arbitrum ($20.5M) — makes it the default, but defaults deserve scrutiny. Panoptic is the wrinkle: LP-based covered calls on Uniswap positions, where impermanent loss does the assignment's job and concentrated liquidity does the strike's. When two-thirds of a category disappears within a year — Stryke, Cega, Ribbon — part of the surviving yield is a risk premium.

Vault versus Deribit, OKX and Bybit

On-chain is boutique: crypto options volume ran $867.5B in H1 2026, up 12% year over year — but only 2.4% of crypto derivatives. Deribit's share fell from 66.2% in 2025 to 41.8% in H1 2026; Bybit is second at 22.4% and leads ETH options with 38% against Deribit's 29%. BTC's DVOL sat at 44.8 on August 24, 2026. All of on-chain is about 5% of premium volume — Derive's 95% share is 95% of a small pie.

For size, exchanges win: deeper books, instant exit, no lockup. The vault buys self-custody between epochs, no KYC, composability — I sell size on an order book and keep the on-chain sleeve small.

The hack that rewired how I vet vaults

On December 12, 2025, Ribbon Finance's legacy dual-option vaults were drained of about $2.7M (roughly $2.3M net after $400K was forfeited). The cause was not exotic: a December 6 oracle upgrade had left transferOwnership and setImplementation unprotected, and the attacker set arbitrary expiry prices for wstETH, AAVE, LINK and WBTC. Ribbon had merged into Aevo in July 2023; every legacy vault was decommissioned afterward, a 19% haircut payout plan drew backlash, and the claims window closed June 12, 2026.

That post-mortem changed my checklist more than any audit report. Premiums pay you for smart-contract risk, oracle risk and admin-key risk — the yield is partially compensation for exactly the thing that failed, the mechanism behind oracle manipulation attacks. Before depositing now I read the upgrade path: who can call setImplementation, is there a timelock, is the price feed multisig-writable. If the docs cannot answer in five minutes, the APY is irrelevant — our how to read audits guide and wallet security checklist cover the rest.

Who should still run this, and how I size it

After 60 days I still run it, with narrower rules. It earns in sideways, high-IV regimes: my six green epochs were a flat tape, and the one red week was a 10% rally. Anchorage's corridor — 10–25 delta, 21-plus day expiries — is where I would sell manually. I cap the sleeve at 10% of my crypto portfolio and treat premiums as a bonus on a position I would hold anyway; if the base position does not make sense, the call premium does not rescue it — the logic in our portfolio allocation guide.

Two suggestions if you try this. First, run the net number before depositing — premiums at your strike and tenor, minus 0.1–0.3% trading fees, minus 0.1% settlement, minus an honest assignment haircut; your realistic annual number likely lands between 8% and 15%, not 30%. Second, decide your exit before entry: epoch-end withdrawals and cooldowns are the rule, so size so a locked position is an inconvenience, not an emergency — check the split with a portfolio allocator.

What the capital could do instead: the baseline is staking ETH three ways; the market-neutral alternative is delta-neutral farming, whose costs live in perpetual funding rate math; proceeds park in stablecoin yield rotation; and a year with Nexus Mutual taught me to price the tail. The one-liner I keep: a covered call vault doesn't pay you for holding ETH — it pays you for selling away the part of ETH you'll miss most. I collected rent for two months, lost the penthouse in week 7, and still think the rent is fair at the right price. If you run a similar ledger, drop your numbers in the comments; the monthly version goes to newsletter readers first, alongside our monthly portfolio review.

Sources and further reading

• Anchorage Digital — Synthetic Yield on Bitcoin — the 5.5% net finding and the 10–25 delta / ≥21-day corridor.
• DefiLlama — Derive Options — TVL $198M and per-chain breakdown (Sep 24, 2026).
• rekt — Aevo/Ribbon legacy vault exploit — the Dec 12, 2025 post-mortem.
• Panoptic — V2: The Vault Suite — Unicorn and PLP vault launch (Jun 29, 2026).
• ChainCatcher — on-chain options market review — H1 2026 volume and venue shares.
• ethereum.org — Layer 2 networks — the Sep 17, 2026 gas figures.

Frequently asked questions

Are DeFi options vaults still profitable in 2026?

Modestly, and only with discipline. My Derive covered-call vault on 8 ETH collected about $823 in premiums (3.8% of notional) across eight weekly epochs in Aug–Sep 2026, then gave back $411 of upside in one assignment, netting roughly 1.9% in 60 days — about 11.5% annualized, not the 30% APY on the vault page. Call selling works best in sideways, high-IV regimes, sized as a small sleeve of a portfolio.

What happens when a covered call vault gets assigned?

Your ETH is sold at the strike price even when the market is higher. In week 7 of my run, ETH closed at $3,050 and my $3,000 strike was assigned: I kept the 0.35% premium but forfeited the $50 gap — about 1.6% of notional — for a net loss versus holding. On Derive and Aevo, settlement is automatic at expiry using the oracle price; nothing to sign, no early roll, and the cash re-enters the next epoch.

What APY do ETH covered call vaults actually pay?

Low-to-mid teens net at best in calm regimes. At ETH near $2,700 with implied vol of 45–60%, 5–10% out-of-the-money weeklies grossed 0.3–0.7% of notional per week — 15–35% annualized before costs. Subtract assignment gaps, 0.1–0.3% trading fees, 0.1% settlement and flat weeks: my realized run was about 1.9% in 60 days, near 11.5% annualized. Anchorage Digital's May 2026 study measured unfiltered 20-delta BTC calls at just 5.5% net over twelve months.

What happened to Ribbon Finance vaults?

Ribbon merged into Aevo in July 2023, and its legacy dual-option vaults were hacked on December 12, 2025. A December 6 oracle upgrade had left transferOwnership and setImplementation unprotected, so an attacker set arbitrary expiry prices for wstETH, AAVE, LINK and WBTC and drained about $2.7M. Every legacy vault was decommissioned, a 19% haircut payout plan drew community backlash, and the claims window ran from December 12, 2025 to June 12, 2026.

Are on-chain options better than Deribit for selling calls?

Different trade-offs, not strictly better. Deribit led a $867.5B H1 2026 crypto options market with 41.8% share, down from 66.2% in 2025, while Bybit took second at 22.4% and leads ETH options with 38%. On-chain venues — about 5% of premium volume, led by Derive — offer self-custody, no KYC and composable vaults, but thinner depth, epoch lockups, oracle and upgrade-key risk, plus 0.1–0.3% fees and 0.1% settlement.

Related: baseline the opportunity cost with staking ETH three ways; price the tail with a year of Nexus Mutual; study the other side of short vol in perpetual funding rate math; and scan the yield traps and red flags before any deposit.